Assessing and COntrolling seizuRe-modulating fluctuatioNs (ACORN)

** 背景 ** 癫痫是最常见和最严重的神经系统疾病之一，其特征是复发性、无端发作，影响全球约5000万人和英国50万患者。在三分之一的患者中，癫痫控制不足，治疗的副作用对生活质量有害。癫痫发作的严重程度差别很大，从轻微的经验现象或不自主运动到癫痫发作伴意识丧失、摔倒在地和全身抽搐。所有癫痫发作都会对生活质量产生不利影响，更严重的形式会带来身体和大脑损伤以及死亡的风险。局灶性癫痫患者通常会经历2-3种类型的癫痫发作，从孤立的经验现象到惊厥性癫痫发作。在某些人中，睡眠不足，压力，月经周期或其他因素可能会导致更严重的癫痫发作。迄今为止，癫痫发作类型的驱动因素似乎是患者特异性的，并且通常没有很好地理解。与临床观察一致，我的研究最近表明，大多数患者从一次癫痫发作到下一次癫痫发作显示出大脑活动模式的变化。有趣的是，这种变异性不是随机的;一些患者的癫痫发作活动模式表现出昼夜节律的波动，而另一些患者则表现出长期的波动。我们的观察也同意越来越多的证据表明，癫痫发作的产生过程似乎遵循每天，每周，每月，甚至季节性波动在大多数患者。此外，还怀疑这些波动还与呼吸和心率等其他生理变化有关，并受到空气温度、湿度和压力等环境因素的影响。** 动机 ** 总之，这些发现表明存在癫痫调节过程，其影响癫痫发作期间的大脑活动模式，进而决定癫痫发作的严重程度。这些代谢调节过程似乎随着时间的推移而波动，因为它们受到其他生理和环境过程的影响，但它们很可能是患者特异性的，目前尚未得到很好的表征或理解。 ** 目标 ** 本研究金的目标是捕捉和分析癫痫调节过程的波动，以预测个体患者癫痫发作的类型或严重程度。我建议使用长期的EEG记录，结合可穿戴传感器和环境传感器数据。在个体患者的水平上，我将首先检查癫痫发作特征，这些特征随时间波动并确定癫痫发作的严重程度，并将它们与生理和环境因素的连续波动相关联。利用这些关联，我将建立并验证癫痫发作严重程度的患者特异性预测模型。最后，我将探索以闭环的方式与一些调节食欲的过程（例如，通过锻炼，饮食或改变药物水平）进行互动。影响 ** 如果成功，这项研究将导致更好地了解癫痫发作活动是如何调制在个别患者。首先，这将产生一个个性化的预测模型，可以预测即将发生的癫痫发作的严重程度。从长远来看，我将开发新的时间适应性治疗策略，通过有针对性和及时地与药物调节过程相互作用，最大限度地减少副作用。这些治疗策略可以是例如药物的自动控制释放，或实时反馈给患者以改变药物剂量或饮食。

**Background**Epilepsy is one of the most common and serious neurological diseases characterised by recurrent, unprovoked seizures affecting around 50 million people globally and 500,000 patients in the UK. In a third of patients, seizure control is inadequate, and the side effects of treatment are detrimental to quality of life. Novel treatment approaches are needed.Seizures vary greatly in their severity, ranging from minor experiential phenomena or involuntary movements through to seizures with loss of awareness, falling to the ground and a generalized convulsion. All seizures adversely affect quality of life and more severe forms carry risk of physical and cerebral injury and of fatality.Individuals with focal epilepsy typically experience 2-3 types of seizures, ranging from experiential phenomena in isolation, to convulsive seizures. In some individuals, lack of sleep, stress, menstrual cycle or other factors may precipitate more severe types of seizures. To date, the factors driving the types of seizure appears to be patient-specific, and generally not well-understood.In agreement with the clinical observations, my research has recently shown that most patients display variability in brain activity patterns from one seizure to the next. Intriguingly, this variability was not random; some patients showed circadian fluctuations in their seizure activity patterns, whilst others showed longer-term fluctuations. Our observations also agree with a growing body of evidence that seizure generating processes appear to follow daily, weekly, monthly, or even seasonal fluctuations in most patients. It is additionally suspected that these fluctuations are also related to other physiological changes such as respiration and heart rate, and influenced by environmental factors such as air temperature, humidity and pressure. **Motivation**Together, these findings suggest that seizure-modulating processes exist, which influence brain activity patterns during a seizure, and in turn determine the severity of a seizure. These seizure-modulating processes appear to fluctuate over time, as they are influenced by other physiological and environmental processes, but they are most likely patient-specific, and currently not well-characterised or understood. **Goal**The goal of this fellowship is to capture and analyse the fluctuations in seizure-modulating processes to predict the type or severity of seizures in individual patients. I propose to do so using long-term EEG recordings, combined with wearable sensors and environmental sensor data. On an individual patient level, I will first characterise key seizure features that fluctuate over time and determine seizure severity, and associate them with the continuous fluctuations in physiological and environmental factors. Using the associations, I will build and validate patient-specific predictive models of seizure severity. Finally, I will explore interacting with some of the seizure-modulating processes (e.g. through exercise, diet or changing medication levels) in a closed-loop manner.**Impact**If successful, this fellowship will lead to a better understanding of how seizure activity is modulated in individual patients. In the first instance, this would result in an individualised predictive model that can forecast the severity of upcoming seizures. In the longer term, I will develop novel time-adaptive treatment strategies which minimise side effects by interacting with seizure-modulating processes in a targeted and timely manner. These treatment strategies could be for example the automated controlled release of a drug, or real-time feedback to the patient to change drug dosage, or diet.